sqoop-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Venkat Ranganathan (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (SQOOP-312) Support for hive dynamic partitions with SQOOP import
Date Tue, 13 Aug 2013 05:46:49 GMT

    [ https://issues.apache.org/jira/browse/SQOOP-312?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=13737854#comment-13737854

Venkat Ranganathan commented on SQOOP-312:

HCatalog integration has been added as part of 1.4.4

A dynamic partition is one where the partitioning keys are specified in Hive and the corresponding
values are coming from DB columns instead of a specific set of values as in static partitioning.
  And the --hive-partitioning-key and value can be used to specify the static partitioning
key as before.

> Support for hive dynamic partitions with SQOOP import
> -----------------------------------------------------
>                 Key: SQOOP-312
>                 URL: https://issues.apache.org/jira/browse/SQOOP-312
>             Project: Sqoop
>          Issue Type: New Feature
>            Reporter: Bejoy KS
> Currently in order to populate hive table dynamic partitions using Sqoop import we need
to perform the following steps.
> 1. Need to analyze the db table and identify the distinct values to be partitioned column

> 2. If there are n distinct values for the column then we need to create n different SQOOP
import commands, each having the corresponding where clause to pick the specific data corresponding
to the value along with --hive-partition-key <key-name/column name> and --hive-partition-value
<value-string/column value>.
> This approach becomes a bottle neck in case of larger tables that spawns millions of
rows. Such tables should be partitioned in hive and there could at lest 300 to 500 partitions,
ie 300 to 500 Sqoop imports. 
> We are currently overcoming this hurdle by the following tweak
> 1. Sqoop import the whole db table into a non partitioned hive table
> 2. Manually create a partition based hive table
> 3. Use hive QL to parse the data from non partitioned hive table to the corresponding
partitions in the partitioned hive table.
> Expecting some parameters in SQOOP import to execute the following within SQOOP itself.

This message is automatically generated by JIRA.
If you think it was sent incorrectly, please contact your JIRA administrators
For more information on JIRA, see: http://www.atlassian.com/software/jira

View raw message